摘要
煤矿安全生产的关键是矿工,而情绪是影响矿工的重要因素,所以有必要对矿工的情绪进行识别。近年来,基于脑电的情绪识别受到了大量的关注,但由于脑电信号微弱,易受干扰,从而降低了情绪识别的精度。针对这一问题,提出优化变分参数与改进小波软阈值重构滤波算法。首先,利用乌燕鸥算法优化变分模态分解的参数,得到一组优化的变分模态分量。接着,通过相关系数差值比的判断条件来区分变分模态的有效分量和含噪分量。然后利用改进的小波软阈值对含噪分量进行分解和重构,得到去噪分量。最后,将去噪分量与有效分量重构,实现所提的滤波算法。结果表明:相比于VMD法、优化参数VMD和小波硬阈值法、优化参数VMD和小波软阈值法,所提滤波算法的信噪比平均提高了3.2847 dB,均方根误差平均降低了0.0695,滤波效果更优。
The key to safe production in coal mine is miners,and emotions are an important factor affecting miners,so it is necessary to identify miners’emotions.In recent years,emotion recognition based on electroencephalograph has attracted a lot of attention,but the EEG signals are weak and easily disturbed,which reduces the accuracy of emotion recognition.Aiming at this problem,a reconstruction filtering algorithm based on optimized variation parameters and improved wavelet soft-threshold is proposed.First,the parameters of the variation mode decomposition are optimized by the sooty tern algorithm,and a group of optimized variation components are obtained.Second,a judgment condition of the difference ratio of the correlation coefficients is used to distinguish the effective components and the noisy components of the variation modes.Third,an improved wavelet soft-threshold is adopted to decompose and reconstruct the noisy components to obtain the denoising components.Fourth,the reconstruction of the denoising components with the effective components is conducted,and the proposed filtering algorithm is completed.The results show that compared to the VMD method,the optimization parameter VMD method with the wavelet hard-threshold,and the optimization parameter VMD method with the wavelet soft-threshold,this filtering algorithm increases the signal-to-noise ratio by 3.2847 dB on average,and decreases the root mean square error by 0.0695 on average,indicating a better filtering effect.
作者
汪梅
王将
李远成
董立红
马天
李铭禹
WANG Mei;WANG Jiang;LI Yuancheng;DONG Lihong;MA Tian;LI Mingyu(College of Computer Science and Technology,Xi’an University of Science and Technology,Xi’an 710054,China;College of Electrical and Control Engineering,Xi’an University of Science and Technology,Xi’an 710054,China)
出处
《西安科技大学学报》
CAS
北大核心
2022年第2期380-388,共9页
Journal of Xi’an University of Science and Technology
基金
国家自然科学基金重点项目(61834005)
中国学位与研究生教育学会项目(B-2017Y1002-170)
陕西省自然科学基金项目(2020JM-525)
榆林市科技计划项目(CXY-2020-026)。
关键词
滤波算法
变分模态
参数优化
阈值函数
小波重构
filtering algorithm
variation mode
parameter optimization
threshold function
wavelet reconstruction
